Knowledge of how mutations affect protein function is important for understanding how natural proteins evolved, engineering new proteins with useful properties, and predicting disease-associated mutations. Epistasis is the phenomenon where the effect of one mutation depends on the presence of another mutation. These epistatic interactions stymie our ability to predict the phenotypic consequences of mutations, and can restrict the mutational pathways that are available to evolution. Laboratory evolution studies have highlighted the multifaceted nature of protein function and how competing biophysical properties, such as enzymatic activity and protein stability, can generate epistatic interactions between residues. I hypothesize that most epistasis arises as a result of these pleiotropic mechanisms, rather than direct physical interactions between residues. The human urea cycle enzyme Arginase I (hArgI) will be used to study how sequence changes affect expression, stability, and enzymatic activity. The experimental approach will leverage recent advances in parallel DNA sequencing and ultra-high- throughput screening to map the protein function landscape on an unprecedented scale. These data will be used to explore how combinations of biophysical properties generate mutational epistasis and define the space of functional protein sequences.